|Home | About | Journals | Submit | Contact Us | Français|
In the mammalian central nervous system, excitatory amino acid transporters (EAATs) are responsible for the clearance of glutamate after synaptic release. This energetically demanding activity is crucial for precise neuronal communication and for maintaining extracellular glutamate concentrations below neurotoxic levels. In addition to their ability to recapture glutamate from the extracellular space, EAATs exhibit a sodium- and glutamate-gated anion conductance. Here we show that substitution of a conserved positively charged residue (Arg-388, hEAAT1) in transmembrane domain 7 with a negatively charged amino acid eliminates the ability of glutamate to further activate the anion conductance. When expressed in oocytes, R388D or R388E mutants show large anion currents that display no further increase in amplitude after application of saturating concentrations of Na+ and glutamate. They also show a substantially reduced transport activity. The mutant transporters appear to exist preferentially in a sodium- and glutamate-independent constitutive open channel state that rarely transitions to complete the transport cycle. In addition, the accessibility of cytoplasmic residues to membrane-permeant modifying reagents supports the idea that this substrate-independent open state correlates with an intermediate outward facing conformation of the transporter. Our data provide additional insights into the mechanism by which substrates gate the anion conductance in EAATs and suggest that in EAAT1, Arg-388 is a critical element for the structural coupling between the substrate translocation and the gating mechanisms of the EAAT-associated anion channel.
In the mammalian central nervous system, excitatory amino acid transporters (EAATs)5 regulate the extent and duration of glutamatergic signaling by efficiently clearing glutamate from the extracellular space into surrounding neurons and glial cells (1). EAATs are secondary active transporters that couple the inward movement of one glutamate molecule to the influx of three sodium ions and one proton and the efflux of one potassium ion (2, 3), resulting in a net influx of two positive charges. In addition, EAATs mediate a thermodynamically uncoupled anion conductance that is gated by sodium and glutamate (4,–7). The EAAT-associated anion conductance has been suggested to promote electrogenic glutamate uptake by clamping the cell at negative potentials (5) and to serve as a glutamate sensor that dampens cell excitability and limits additional glutamate release (8, 9).
The structural basis for substrate translocation by EAATs is not fully understood. However, our understanding of the transport mechanism has been significantly expanded by both functional studies of mammalian carriers (see Ref. 10 for a review) and crystallographic data on the structure of an archaeal ortholog from Pyrococcus horikoshii (GltPh) (11,–15). Both prokaryotic and mammalian isoforms assemble as homotrimeric complexes consisting of three identical subunits (11, 16,–18). Each subunit contains eight transmembrane domains (TMs) and two reentrant hairpin loops, HP1 and HP2 (11, 19,–23) (Fig. 1A). The first six TMs of each protomer participate in intersubunit contacts (11,–13), with TM2, -4, and -5 serving as the interaction surface. These intersubunit contacts form a scaffold that surrounds a highly conserved compact core domain comprising HP1, TM7, HP2, and TM8 (13). HP1 and HP2 have been proposed as the internal and the external gates of the transport mechanism, respectively (12, 24). The opening of HP2 exposes the substrate binding site (12, 24), and after substrate and co-transported ions bind, HP2 closes and occludes the binding sites. After HP2 closure, the transport domain has been proposed to move in a piston-like motion toward the cytoplasm, which enables the opening of the internal gate, HP1 (12). The subsequent binding of potassium ions facilitates the reorientation of the transporter from inward to outward facing states to complete the cycle.
Much less is known about the structural elements that mediate the uncoupled anion conductance. The EAAT-associated anion conductance is a pore-mediated anion permeation process (6) with two predominant conducting states: a sodium-activated conductance (6, 25,–27), commonly called leak conductance, and a substrate-gated conductance, which also requires the presence of Na+ (4, 5, 25). Several residues located on TM2 can affect anion permeation and selectivity (28), suggesting that this region may be part of the anion permeation pore. In addition, other conserved residues in TM5, TM7 (29), and HP2 (30) contribute to anion permeation without affecting substrate transport (28,–30). The “gating,” or opening and closing of the EAAT-associated anion channel, depends on binding of the substrates, Na+ and glutamate, and thus structural elements must exist that enable coupling between the substrate-bound states and anion channel opening. Although such a coupling has been suggested previously (31,–34), the structural elements responsible for this coupling have not been identified.
Here, we show that substitution of a highly conserved basic residue in TM7 (Arg-388 in EAAT1) with an acidic amino acid (Asp or Glu) disrupts anion channel gating by causing the channel to reside preferentially in a constitutively open state. In these two mutants, the anion conductance is active in the absence of substrate and shows no further enhancement in the presence of saturating concentrations of sodium and glutamate. Additionally, these mutations severely disrupt glutamate transport, most likely because they spend more time in the anion conducting state. Both anion permeation and substrate transport remain sensitive to the non-transported substrate analog dl-threo-β-benzyloxyaspartate (dl-TBOA), demonstrating that the substrate binding site remains accessible. Experiments assessing the accessibility of introduced cysteine residues support the idea that the mutated carriers favor a conformation resembling an intermediate outward facing state (iOFS) (14), which reflects a partial transition of the transporter from the outward facing state (13, 14). Our results support the hypothesis that the transition to an iOFS may be closely linked to anion channel gating (14, 35) and suggests a structural coupling between substrate translocation and channel activation that depends upon the presence of a positive charge at Arg-388.
Point mutations were introduced into the EAAT1 WT, EAAT1 WT cysteineless, and EAAT4 WT cDNAs using the QuikChange mutagenesis kit (Stratagene). EAAT1 and EAAT4 capped cRNA were synthesized from SmaI-linearized pOTV through use of MESSAGE machine kits (Ambion, Austin, TX). Subsequently, cRNA was resuspended in 10 μl of water and then adjusted to a concentration of ~300 μg/ml and stored in 2-μl aliquots at −80 °C until use. 50 nl of ~300 μg/ml cRNA (unless stated otherwise) were injected into oocytes using a Nanoliter 2000 injector (World Precision Instruments, Sarasota, FL), and oocytes were kept at 18 °C in ND96 solution supplemented with 2.5 mm sodium pyruvate and 100 μg/ml gentamycin sulfate prior to recordings. Oocytes expressing EAAT1 and EAAT4 were measured 2–3 days after injection.
EAAT-associated currents were recorded in oocytes by two-electrode voltage clamp using a GeneClamp500B (Axon Instruments, Union City, CA) amplifier. NO3− was used as a main permeant anion. Recording solutions contained 96 mm NaNO3, 4 mm KCl, 0.3 mm CaCl2, 1 mm MgCl2, 5 mm HEPES, pH 7.4, in the absence or in the presence of 1 mm external glutamate. Oocytes were held at −60 mV, and currents were elicited by 200-ms voltage steps between −100 and +60 mV, filtered at 2 kHz and digitalized with a sampling rate of 10 kHz using a Digidata AD/DA converter (Axon Instruments). The current-voltage relationship curves were plotted without using any current subtraction procedure. To determine the IC50 for dl-TBOA, we measured the current amplitudes at several concentrations of dl-TBOA (0.1, 1, 5, 10, 30, 50, 100, and 500 μm) and then fit the curves with a four-parameter logistic curve to obtain the IC50.
[3H]Glutamate uptake assays were performed on Xenopus oocytes. The uptake buffer contained 96 mm NaNO3, 4 mm KCl, 0.3 mm CaCl2, 1 mm MgCl2, 5 mm HEPES, pH 7.4. Right before the experiment, labeled (500 nm) and unlabeled (1 mm) glutamate was added to reach saturating glutamate concentrations. Oocytes were incubated in radioactive buffer for 10 min and then transferred three times into ice-cold ND96 containing 96 mm NaCl, 4 mm KCl, 0.3 mm CaCl2, 1 mm MgCl2, 5 mm HEPES, pH 7.4, to stop transport activity. Immediately after, oocytes were individually transferred into scintillation counting vials containing 0.4 ml of 1% SDS. After gently shaking the vials for 1 h, scintillation solution was added, and the samples were counted.
For cysteine modification experiments, oocytes expressing WT (cysteineless) and mutant transporters were incubated for 5 min in ND96 buffer containing different sulfhydryl-modifying reagents and then washed with ND96 before proceeding with the experiment. N-Ethylmaleimide (NEM) was diluted in EtOH to 100 mm, divided into aliquots, and frozen for up to 1 month. Fresh stock solution of 10 mm NEM in ND96 was prepared for every experiment, and a final concentration of 100 μm NEM was used in the modification experiments. [2- (trimethylammonium)ethyl]methanethiosulfonate (MTSET) was diluted in water to 1 m, divided into aliquots, and frozen for up to 6 months. MTSET was added to the experimental solution to a final concentration of 1 mm.
Forty-eight hours after mRNA injection, equal groups of 6–8 oocytes were washed three times with ice-cold ND96 solution and equilibrated to 4 °C for 10 min. Then the groups were incubated with gentle agitation for 30 min at 4 °C with 1 ml of 1.5 mg/ml EZ-LinkTM Sulfo-NHS-SS-Biotin (Thermo Scientific) prepared in biotinylation buffer (150 mm NaCl, 2 mm CaCl2, 10 mm triethanolamine, pH 7.8). The reaction was quenched by incubating the oocytes for 15 min with 50 mm glycine in ND96 solution. Oocytes were washed with ND96 and lysed in radioimmune precipitation buffer (10 mm Tris, 150 mm NaCl, 1 mm EDTA, 0.1% SDS, 1% Triton X-100, and 1% sodium deoxycholate, pH 7.4) at 4 °C for 1 h, followed by two sequential centrifugations (500 × g and 15,000 × g). The cleared supernatant was divided into two aliquots. One aliquot, the surface fraction, was used for isolation of biotinylated proteins with NeutrAvidin®Plus UltraLink resin (Thermo Scientific). The second aliquot, the input fraction, was used to determine total protein (BCA, Thermo Scientific) and total EAAT levels. Samples were loaded to 10% TGX Stain-FreeTM gel (Bio-Rad) and separated by SDS-PAGE. After the electrophoretic transfer to LF-PVDF membrane, the membrane was activated with UV to visualize total protein and blotted with the anti-EAAT1 antibody (EAAT1-2510, rabbit polyclonal antibody produced in house) and Dylight649-conjugated anti-rabbit secondary antibody (Jackson ImmunoResearch). Images were obtained with a ChemiDoc MP Imager, and analysis of bands was performed with ImageLab version 5.0 software (Bio-Rad).
Two-electrode voltage clamp data were analyzed using pClamp9 (Axon Instruments), and the results from the electrophysiological recordings, biochemical experiments, and transport assays were analyzed using SigmaPlot (Jandel Scientific, San Rafael, CA). For statistical evaluation, Student's t test was used.
EAAT1 Arg-388 is a conserved positively charged residue located at the cytoplasmic end of TM7 (Fig. 1, A and B), in the vicinity of other residues reported to affect anion channel function in EAATs (29). To investigate the effect of this residue on anion permeation, Arg-388 was mutated to either aspartate (EAAT1 R388D) or glutamate (EAAT1 R388E). The EAAT-associated anion channels display a permeability sequence of SCN− > ClO4− > NO3− > I− > Br− > Cl− > F− (36). Using NO3− as a main permeant anion, we measured current-voltage relationships in response to a family of voltage pulses from −100 to +60 mV in the absence and the presence of 1 mm l-glutamate from oocytes expressing EAAT1 WT, EAAT1 R388D, or EAAT1 R388E (Fig. 1, C–E, respectively). As shown in Fig. 1, oocytes expressing EAAT1 WT in the absence of glutamate displayed an averaged macroscopic current amplitude of 0.8 ± 0.1 μA at +60 mV (Fig. 1C, black closed circles, n = 18). Upon application of 1 mm glutamate, macroscopic current amplitudes were increased by 3-fold, resulting in an average current amplitude of 2.1 ± 0.2 μA at +60 mV (Fig. 1C, red open circles). In contrast, in oocytes expressing EAAT1 R388D or R388E mutant transporters, the macroscopic current remained unchanged after application of 1 mm glutamate. The average macroscopic current amplitudes obtained at +60 mV from oocytes expressing EAAT1 R388D were 0.9 ± 0.1 and 0.9 ± 0.1 μA in the absence and the presence of glutamate, respectively (Fig. 1D; n = 18). Similar results were obtained from oocytes expressing EAAT1 R388E, which had current amplitudes of 1.0 ± 0.1 μA in the absence and 1.0 ± 0.1 μA in the presence of 1 mm glutamate (Fig. 1E; n = 17). Saturating concentrations of other known substrates, such as d-aspartate or l-cysteine, also failed to stimulate additional currents in oocytes expressing either of the mutant carriers (Fig. 1, G and H).
All currents measured in wild type- and mutant-expressing oocytes were at least 5-fold larger than the currents measured in water-injected oocytes in the same ionic conditions (0.2 ± 0.01 μA at +60 mV; see Fig. 1, C–H), indicating that the substrate-insensitive currents in oocytes expressing the mutant transporters were predominantly EAAT-mediated. To further confirm that the substrate-independent currents were EAAT1-mediated, we perfused the oocytes with different concentrations of dl-TBOA, a well described inhibitor of EAAT transport and channel activity. Upon application of 30 μm dl-TBOA, we observed a 30% inhibition of anion currents in oocytes expressing the mutant transporters (Fig. 1, G and H, light gray bars). The IC50 for inhibition of the anion currents by dl-TBOA was substantially higher for both mutant transporters (50.7 ± 3.3 μm for R388D (n = 4) and 43.7 ± 3.2 for R388E (n = 5)) than for the WT EAAT1 (7.4 ± 1.8, n = 5). Using a higher concentration of dl-TBOA (500 μm) to ensure maximum blockade, all currents obtained from oocytes expressing EAAT1 R388D or EAAT1 R388E mutant transporters were reduced to background levels, resulting in average currents at +60 mV of 0.4 ± 0.04 μA and 0.3 ± 0.04 μA, respectively (Fig. 1, F–H, dark gray bars; n > 5). No additional current was observed in oocytes expressing either mutant transporter at any glutamate concentration tested (up to 1 mm; Fig. 1I). Together, these data confirm that the substrate-insensitive anion currents measured in oocytes expressing R388D or R388E are indeed EAAT-mediated currents. In addition, the data strongly suggest that substitution of Arg-388 with a negatively charged amino acid disrupts the gating mechanism of the anion channel.
Our observation of large, dl-TBOA-blockable, substrate-independent anion currents in oocytes expressing R388D or R388E supports the idea that the mutants reside at the plasma membrane. To corroborate the surface expression of the mutants, we conducted surface biotinylation assays in oocytes expressing WT or mutant carriers in four independent batches, all showing comparable anion current amplitudes. Analyses of data from the biotinylation experiments show that the cell surface expression of both EAAT1 R388D and EAAT1 R388E was reduced by more than 90% when compared with wild type transporters (Fig. 2, A and C). In addition, the absolute expression of the mutant transporters obtained from the total oocyte extract was also reduced by more than 90% (Fig. 2B). Irrespective of expression levels, the ratio of surface-expressed to total transporter protein expression was comparable for wild type and mutant carriers (Fig. 2D), indicating that the mutations do not modify the efficiency of trafficking to the plasma membrane. The reduced expression of the mutant transporters was consistently observed using different batches of RNA and may reflect alterations in the efficiency of biosynthesis of the mutated carriers.
To assess the function of the mutant carriers compared with WT, it was important to measure substrate transport in oocytes expressing similar numbers of carriers at the cell surface. In our experiments, we normally used an RNA concentration of 300 μg/ml. Because we observed substantially greater total expression of WT compared with R388D and R388E mutants, we diluted the WT RNA to achieve expression levels more comparable with those of the mutants. Increasing dilutions of the hEAAT1 WT RNA resulted in a corresponding decrease in surface expression (Fig. 2E), transport currents (dark gray bars), glutamate uptake (light gray bars), and anion currents (white dashed bars) (Fig. 2F). We then used a 20 μg/ml dilution of EAAT1 WT RNA, which yielded an expression level more comparable with those of the mutant carriers (Fig. 2, G–I).
Substrate transport measurements provide a functional indicator of the surface expression of glutamate transporters. To investigate whether the mutants display altered transport activity, we measured radiolabeled glutamate uptake using saturating glutamate concentration (1 mm final) supplemented with 500 nm [3H]glutamate. After normalizing to surface expression, low (<10%; Fig. 3B) but significant sodium-dependent, dl-TBOA-sensitive transport activity was observed with both EAAT1 R388D and EAAT1 R388E (Fig. 3C), indicating that although transport is impaired, the substrate binding site remains accessible in the R388D and R388E mutant carriers.
To corroborate these findings, we attempted to measure transport currents in oocytes expressing WT and mutant transporters. The current measured at negative potentials after application of glutamate to EAAT-expressing cells predominantly reflects the coupled movement of glutamate, sodium, protons, and potassium across the plasma membrane and is commonly referred to as the transport current (25, 37). In oocytes expressing EAAT1 WT (20 μg/ml), we observed transport current amplitudes of −229.5 ± 23.7 nA (n = 18) at −100 mV (data not shown). However, these currents were not detected in oocytes expressing R388D or R388E because the current amplitudes were not different from those of water-injected oocytes (−12.4 ± 4.2 nA (n = 16) for R388D, −6.8 ± 2.5 nA (n = 17) for R388E, and −7.2 ± 1.8 nA (n = 16) for water-injected). Given that oocytes expressing R388D or R388E displayed uptake activities of 2.4 ± 0.6% (n = 27) and 3.3 ± 0.8% (n = 16) relative to wild type, respectively (Fig. 3A), then, based on ion stoichiometry, transport current amplitudes for the mutants would be predicted to be smaller than 7 nA, a value that is within the background of the system. The data presented thus far demonstrate that the R388D and R388E mutants can reach the cell surface and do retain some capacity to bind and translocate glutamate; however, their transport activity is severely impaired, and they mediate a larger anion conductance that cannot be further activated by substrates.
The reduced surface expression seen in both mutant carriers could explain the absence of a current increase in the presence of glutamate; however, closer scrutiny of the data makes this possibility unlikely. As shown in Fig. 1, the mutant carriers showed measurable macroscopic anion currents in the absence of glutamate (sodium-activated leak current). In transporters with a normal substrate-gated anion conductance, application of saturating concentrations of glutamate would be expected to result in a 2–3-fold increase in the current amplitude. This has been consistently observed for EAAT2 (6), EAAT3 (37), EAAT4 (6, 9, 30), and EAAT5 (38). As predicted, oocytes expressing EAAT1 WT displayed current amplitudes in the presence of 1 mm glutamate that were 3-fold larger than in its absence (Fig. 1C). Moreover, in oocytes injected with EAAT1 WT RNA diluted to 20 μg/ml, we observed a 3-fold increase in the macroscopic current amplitude upon application of glutamate (Fig. 4A). Even with EAAT1 WT diluted to 7 μg/ml, which resulted in a ~50% lower surface expression than R388D (Fig. 2G), we still observed an ~3-fold increase in current amplitude with glutamate application (61.6 ± 16.8 nA in the absence and 153.2 ± 25.7 nA (n = 6) in the presence of glutamate; data not shown). These data illustrate that even when very few carriers are expressed in the plasma membrane, glutamate elicits a 2–3-fold increase in current amplitude, thus demonstrating that reduced expression does not explain the absence of glutamate-activated conductance in the mutant carriers.
The inability of saturating concentrations of substrate to increase the macroscopic current in the R388D and R388E mutants suggests that the mutation may alter the conformation in a way that drives the protein into a constitutively open channel state. The large currents displayed by the mutants suggest the existence of a constitutive open channel. After normalizing the maximum current amplitudes (current in the presence of 1 mm glutamate) to surface expression, we observed currents that, in both mutants, were more than 6-fold larger than the maximum currents measured in oocytes expressing WT (20 μg/ml) (Fig. 4B, dark gray bars). Substitution with another permeant anion, SCN− (Fig. 4B, light gray bars) also resulted in greater than 6-fold larger currents that in WT (20 μg/ml).
EAAT-associated anion channels also exhibit voltage-dependent gating (6, 37,–40). To further investigate whether EAAT1 R388D exists in a constitutive open state, we measured the voltage dependence of the relative open probability (see Ref. 37). We applied a family of voltage pulses from −100 to +60 mV, followed by a test pulse at +40 mV. We next measured the current amplitude at the test pulse (Fig. 4C, red dashed line) and plotted it against the previous potential. The instantaneous current amplitude at the test pulse depends on the number of channels in the membrane (N), the single channel current amplitude at that voltage (i), and the absolute open probability at the end of the preceding potential (p) (I = iNp) (41). Assuming no change in the open probability from the voltage step from the preceding potential to the test potential, the normalized instantaneous current at the test potential is proportional to the open probability at the end of the prepulse and thus reflects the voltage dependence of the relative open probability of the channel. As shown in Fig. 4, C and D, the relative open probability of EAAT1 WT displays a voltage dependence typically observed in members of this carrier family. However, no changes in relative open probability were observed from −100 to +60 mV, with EAAT1 R388D (Fig. 4, C and D) or EAAT1 R388E (data not shown), further suggesting that in the mutant carriers, the anion channels preferentially exist in an open conformation that rarely transitions to a closed state.
To ensure that the larger substrate-insensitive currents do not result from alterations in the structure of the permeation pathway caused by the mutation, we measured relative current amplitudes and permeability ratios (6, 28, 30) in EAAT1 WT and EAAT1 R388D transporters. The relative current amplitudes for the different permeant anions at +60 mV were similar for both WT and R388D (Fig. 5, A and B). To estimate the relative permeability ratios, we measured the reversal potential for the different anions and calculated the ratios using the Goldman-Hodgkin-Katz equation (42). The permeability ratios were similar for WT and mutant transporters (Fig. 5C). Both of these results suggest that the integrity of the permeation pathway is not affected by the Arg-388 point mutation.
One could also argue that the large constitutive currents observed in the mutant carriers could be toxic to the oocytes or that there might be an upper limit on the macroscopic current amplitudes that can be sustained by the oocytes. To rule out these possibilities, we measured the macroscopic current amplitudes 3 and 4 days after injection, when we observed currents with the larger amplitudes. Four days after injection of EAAT1 R388D into oocytes, we obtained large current amplitudes in NO3-containing buffers (2.8 ± 0.5 μA in the absence and 2.8 ± 0.6 μA in the presence of glutamate at +60 mV (n = 3)) that remained insensitive to glutamate (data not shown). In addition, inclusion of 500 μm dl-TBOA in the oocyte incubation medium from the time of injection until transfer to the recording chamber had no effect on the maximum current amplitudes. Using NO3− as the main permeant anion, current amplitudes at +60 mV were 2.1 ± 0.6 μA (n = 6) in dl-TBOA-incubated oocytes and 2.2 ± 0.8 μA (n = 5) without dl-TBOA. Moreover, the similar reversal potentials observed in oocytes expressing WT and mutant carriers provide further evidence confirming the viability of the oocytes.
The EAAT-associated anion conductance is also sodium-dependent (25,–27, 43, 44). Thus, if the mutant carriers exist in a constitutively open state, one would predict that anion permeation would persist in the absence of extracellular Na+. To explore this possibility, we measured the sodium dependence of anion channel activation. Using choline-NO3 as a substitute for NaNO3, we assessed the anion conductance at different sodium concentrations. The anion current at +60 mV in oocytes expressing EAAT1 WT was clearly Na+-dependent (Fig. 5D, open circles). However, in oocytes expressing EAAT1 R388D or EAAT1 R388E, complete substitution of extracellular sodium by choline did not alter the amplitude of the currents, which were greater than 98% of the current at saturating sodium concentrations (96 mm Na+) (Fig. 5D, open squares and open diamonds). To rule out the possibility that the sodium independence of activation of the anion channel was caused by impairment of sodium binding, we measured substrate transport at different sodium concentrations. As observed with the WT carrier, transport by the R388D mutant remained sodium-dependent (Fig. 5D, inset). These data provide further support for the idea that substitution of the conserved Arg-388 by a negatively charged amino acid results in channels that reach their maximum open probability even in the absence of glutamate and/or sodium.
To examine whether mutations at this position in other isoforms produce a similar phenotype, we substituted a glutamate residue at the corresponding position in the neuronal isoform, EAAT4 (EAAT4 R410E) and measured currents and glutamate uptake in either the mutant or wild type EAAT4. In oocytes expressing EAAT4 WT, the average current amplitude in an NO3-based solution at +60 mV was 0.5 ± 0.01 μA in the absence of glutamate (Fig. 6A, left, black closed circles). These currents were increased up to 3-fold upon application of 1 mm glutamate, resulting in an average macroscopic current amplitude of 1.8 ± 0.03 μA (Fig. 6A, left, red open circles). As seen with R388D and R388E mutations in EAAT1, there was no change in the macroscopic current amplitudes measured in the absence (1.1 ± 0.02 μA; black closed triangles) or presence of 1 mm glutamate (1.1 ± 0.02 μA; red open triangles) in oocytes expressing EAAT4 R410E (Fig. 6A, right). For all three mutants (R410E in EAAT4 and R388D and R388E in EAAT1), the inability of glutamate to elicit additional anion current was observed across the entire voltage range. We then measured the sodium dependence of substrate-gated currents mediated by EAAT4 WT and EAAT4 R410E and found that, as observed with EAAT1 WT, EAAT4 WT currents were fully dependent on extracellular sodium. EAAT4 WT anion currents were reduced to background levels with complete choline substitutions, as observed previously (30, 43). In contrast, after complete substitution of extracellular sodium by choline, the currents measured in oocytes expressing EAAT4 R410E were similar in magnitude (>85%) to those obtained in the presence of saturating sodium concentrations (Fig. 6B). We also analyzed the relative current amplitudes and permeability ratios in EAAT4 WT and R410E transporters, and, as observed with EAAT1, the permeation properties were not affected by the mutation (Fig. 6, C and D). As observed for EAAT1 WT and EAAT4 WT, substrate transport by EAAT4 R410E remains sodium-dependent (Fig. 6B, inset). Thus, in multiple EAAT isoforms, a positively charged residue at the tip of TM7 appears critical for gating of the substrate-gated anion conductance.
To explore the phenotypes produced by other amino acid substitutions, we assessed transport activity and surface expression and measured sodium- and substrate-activated anion currents for mutants with lysine, asparagine, glutamine, and alanine residues at Arg-388. Substitution of Arg-388 with another positively charged residue, lysine (R388K), did not alter transport activity or anion channel gating (Fig. 7). When compared with WT EAAT1, the mutant transporters with asparagine (R388N), glutamine (R388Q), or alanine (R388A) substitutions display slight increases in sodium-dependent currents (Table 1; see normalized macroscopic current amplitudes) and modest reductions in substrate transport (Fig. 7B) and glutamate-gated anion currents (Fig. 7C). However, in contrast to the R388D/E mutants, anion permeation was still enhanced by glutamate in the R388K/N/Q/A substitution mutants (Fig. 7C). These data suggest that only negatively charged residues introduced at this position drive the channel into the substrate-insensitive constitutively open state.
Glutamate transporters can exist in at least two structurally distinct conformations, outward and inward facing. During substrate translocation, carriers are thought to transition between these two orientations through the movement of a large core domain of the protein, referred to as the transport domain (11, 13). Based on high resolution structural data from GltPh (14) and the combination of electrophysiological data, kinetic modeling simulations (32), and voltage-clamp fluorometry (33), it has been speculated that the movement of the transport domain might be coupled to the gating of the substrate-gated anion conductance. However, the relationship between these conformational changes and the mechanism for gating of the anion conductance by substrates remains unclear. Because the anion channels associated with EAAT1 R388D and R388E are no longer sensitive to substrate and appear to favor a constitutively open state, we next tested whether these mutations might be interfering with the movement of the transport domain. To test this hypothesis, we introduced reactive cysteines into the mutant EAAT1 R388D, which had been created within a cysteineless EAAT1 parent construct. We then examined the accessibility of these residues to thiol-modifying reagents, an approach commonly used to predict transmembrane topology and protein conformations (21, 45, 46). We have previously confirmed that the phenotypes of EAAT1 WT and EAAT1 cysteineless are indistinguishable in assays of surface expression, substrate transport, or anion permeation (47, 48). We focused on L376C, a conformationally sensitive residue that has been shown to be modified by the membrane-permeable reagent NEM only when the transporter is driven into an inward facing conformation in the presence of high extracellular K+ (45, 47). Cysteine residues exposed to the cytoplasm can be modified by NEM by reacting irreversibly with the deprotonated form of the cysteine sulfhydryl group. The single cysteine mutant EAAT1 L376C displayed a normal current response to glutamate (Fig. 8A) (for comparison, see EAAT1 WT (Fig. 1C)). As observed with the single R388D mutant (Fig. 1D), oocytes expressing the double mutant (EAAT1 L376C/R388D) showed large constitutive anion currents that did not increase after application of saturating concentrations of glutamate (Fig. 8B).
We next examined the accessibility of the introduced cysteine residue both in the parent L376C and in the L376C/R388D double mutant. Under normal sodium conditions, where the transporter spends most of the time in the outward facing conformation, EAAT1 L376C was not accessible to NEM (Fig. 8C). However, when EAAT1 L376C carriers were exposed to a high K+ extracellular solution, which favors inward facing conformations, L376C became accessible to modification by 100 μm NEM, reducing glutamate uptake by 35% (Fig. 8C) (47). We then performed the modification experiment in cells expressing the L376C/R388D double mutant. A brief incubation of cells expressing the double mutant with 100 μm NEM under physiological conditions (96 mm sodium-based solution), was sufficient to cause an ~50% reduction in transport activity when compared with control conditions (Fig. 8D), which stands in contrast to the absence of inhibition observed with the Leu-376 single mutant (Fig. 8C). Moreover, in the high extracellular K+ condition, the transport activity in cells expressing L376C was only reduced by 35% after NEM modification, whereas the reduction in transport activity in cells expressing L376C/R388D was greater than 90% (Fig. 8D). These data indicate that NEM modifies the carrier from within the cell and that in the double mutant, L376C displays greater cytoplasmic accessibility even under conditions that normally favor outward facing conformations.
The accessibility of L376C in the presence of Na+ in the double mutant could be a consequence of a structural disruption in the R388D mutant that makes L376C accessible even in outward facing conformations. To rule out this possibility, we first tested the effect of NEM in oocytes previously incubated with 500 μm dl-TBOA, which would trap the transporter in an outward facing conformation (12). Preincubation with dl-TBOA prevented the reduction in uptake caused by the NEM modification (Fig. 8F), supporting the idea that L376C is not accessible from the extracellular milieu. We also attempted to modify L376C with MTSET, an impermeant thiol reagent that readily modifies and reduces the uptake activity of several extracellular cysteine substitution mutants in EAAT1 (45, 48). Incubation with 1 mm MTSET did not affect uptake in oocytes expressing the single (L376C) or the double (L376C/R388D) mutant (Fig. 8, G and H). This experiment was done in parallel with oocytes expressing EAAT1 V449C, a mutant known to be inhibited by modifications with MTSET (47, 48), which served as a positive control. Oocytes expressing EAAT1 V449C showed an ~80% reduction in the uptake activity after a 5-min incubation with 1 mm MTSET (also seen in Fig. 9C) (48). These results indicate that the modification observed after incubation with 100 μm NEM in oocytes expressing the double mutant L376C/R388D was caused by exposure of the cysteine residue at the intracellular side of the membrane. From these results, we conclude that when a negatively charged residue occupies Arg-388, L376C is more exposed to the intracellular environment, consistent with a shift of the transport domain toward the cytoplasm.
Our data suggest that the substitution of Arg-388 by a negatively charged residue leads to structural changes that cause the channel to be constitutively open. To further address this issue, we used cysteine modification to restrict the carrier to a more outwardly oriented conformation. Modification of a cysteine substitution in the extracellular gate, HP2, of EAAT1 (V449C) by MTSET has been shown to prevent transition of the carrier to inward facing conformational states (47), and as a result, substrate transport is inhibited, but anion channel gating and permeation are preserved (48). We generated the double mutant V449C/R388D and examined the ability of glutamate to gate the anion conductance. In contrast to the robust glutamate-gated currents observed with the modified V449C single mutant (Fig. 9A, gray bars), the R388D/V449C mutant displayed no glutamate-gated currents (Fig. 9B). The effects of modification on transport by both single and double mutants are completely reversed by DTT (Fig. 9, C and D), confirming that the modification occurs in both mutants. We also observed a modest increase in anion currents after modification with MTSET, a finding consistent with previous studies of V449C and several other phenotypically similar mutants, where modifications disrupt substrate transport but preserve the substrate-gated anion conductance. These observations indicate that even under conditions where the global conformation is restricted to a state where gating (but not transport) readily occurs in the single mutant (V449C), substrate-dependent anion channel gating is still disrupted in the double mutant (R388D/V449C). Thus, even under conditions where the carrier is restricted to conformational states where anion channel gating is favored, the mutant remains selectively stabilized in an open channel state.
In ligand-gated ion channels, pore opening is energetically coupled to the binding of an ion or ligand, which triggers the conformational change required to open the channel. This mechanism for gating is generally governed by a region within the protein that is distinct from the ion permeation pathway (49). However, chloride channels (ClCs) display a different gating mechanism. For the fast gate of ClCs, the gating domain and the anion pore are intimately coupled. In the closed state, the side chain of a glutamate residue occludes the anion pore, and channel opening occurs when conducting anions repel the negatively charged carboxyl group (50, 51). Neutralizing this glutamate residue causes the conduction pore to remain open most of the time (51). Thus, the permeating anion controls the open probability by altering the local motions of the side chain of a gating residue that is within the permeation pathway (51, 52).
A gating mechanism similar to that of ClCs has been proposed for EAATs (28). The substitution of an aspartate residue with alanine (D112A in the TM2-TM3 loop of EAAT1) altered anion permeation by significantly increasing the leak conductance ~3-fold and decreasing the l-glutamate- and d-aspartate-gated conductance by more than 50% (28). It has been proposed that Asp-112 may form part of the gate of the anion permeation pathway, similar to the role of the acidic residue in ClCs (51). Both the intracellular and extracellular anion compositions affect the time and voltage dependence of EAAT-associated anion channel gating (6, 30, 39), suggesting that chloride itself might participate in the activation of the anion conductance in a manner similar to ClCs (52, 53). However, in the EAATs, unlike the fast gate of ClCs, glutamate and sodium binding are also required for channel activation, suggesting that there is a more complex process that links substrate- and ion-dependent conformational changes to enable channel opening.
Here we identify a point mutation that significantly modulates the EAAT-associated anion channel and provides insight into the conformations required for channel gating. Arg-388 is a highly conserved residue located near the cytoplasmic end of TM7, in close proximity to the tips of HP1 and HP2 and to the NMDG motif, the region that comprises the substrate binding pocket (10, 11) (Fig. 1, A and B). The presence of a positive charge at Arg-388 is clearly important. Substitution of Arg-388 with a small neutral alanine (EAAT1 R388A) resulted in ~60–70% reduction in both the transport rate and the glutamate-gated anion conductance (Fig. 7). Similar phenotypes were obtained with substitution of asparagine or glutamine, which are polar, uncharged, and similar in structure to aspartate and glutamate, respectively. However, a more dramatic effect was observed when we reversed the charge by mutating Arg-388 to aspartate or glutamate (EAAT1 R388D or EAAT1 R388E). The sodium-activated leak conductance in EAAT1 R388D and EAAT1 R388E was much larger than in EAAT1 WT-expressing oocytes. Moreover, no additional current was observed following application of glutamate up to 1 mm (Fig. 1I) or at saturating concentrations of alternative substrates, such as aspartate or cysteine (Fig. 1, F–H). The inability of substrates to further increase the macroscopic current amplitudes in the R388D/E carriers could be the result of alterations in the glutamate binding site; however, this possibility is unlikely because the mutant transporters can still accumulate glutamate (although at much reduced rates, ~6% of WT; Figs. 3B and and77B), and glutamate transport can be inhibited effectively by dl-TBOA (Fig. 3C). Both these findings are consistent with the idea that the substrate binding site remains largely intact.
The larger maximum current amplitudes displayed by EAAT1 R388D and EAAT1 R388E (Fig. 4B and Table 1), together with the inability of substrate to further activate the anion conductance, strongly suggest that the anion channel exists in a constitutively open state. The activation of EAAT-associated anion channels has also been observed to be voltage-dependent (6, 37,–40). In oocytes expressing EAAT1 WT, we observed a voltage-dependent relative open probability with a V50 of −25.6 ± 1.2 mV (Fig. 4, C and D). However, no changes in open probability were observed from −100 to +60 mV in oocytes expressing EAAT1 R388D (Fig. 4, C and D) or R388E (data not shown), because the macroscopic currents remained at maximum amplitude throughout the entire voltage range. Additional experiments demonstrated that the anion channel remains open even in the absence of sodium. In both R388E and R388D EAAT1 mutants (Fig. 5D) and in the corresponding mutant R410E in EAAT4 (Fig. 6B), complete substitution of choline for sodium had little effect on the amplitude of anion currents, further supporting the idea that the mutations restrict channel closure and stabilize an open channel state. Arg-388 is in close proximity with several of the residues proposed to be part of the three sodium binding sites (for a review, see Ref. 10). The sodium independence of channel gating might suggest that the mutation disrupts one or more sodium ion binding sites. However, in oocytes expressing EAAT1 R388D or R388E or EAAT4 R410E, glutamate uptake was eliminated by complete substitution of sodium by choline (Figs. 3C, ,55D (inset), and and66B (inset)). These data strongly suggest that sodium binding remains largely preserved in the mutant carriers. Thus, our experiments suggest that when a negative charge occupies position Arg-388 in EAAT1 or Arg-410 in EAAT4, the anion channel exists in a constitutive open state that no longer depends on the substrate and co-transported ions that are normally required to activate the channel (4, 5, 25). These results underscore the relevance of Arg-388 to the gating mechanism of the EAAT-associated anion channel and suggest that the residue may be one of several structural elements that control the coupling between substrate translocation and anion channel gating. We speculate that Arg-388 may be coordinating channel opening and closing by interacting with other residues in its vicinity.
Several high resolution crystal structures have been resolved for the archaeal glutamate transporter ortholog GltPh (11,–14). Intriguingly, the archaeal carrier has also been proposed to conduct anions in a substrate-dependent manner similar to their eukaryotic counterparts (54). The available crystallographic data demonstrate that GltPh can exist in an outward facing conformation, with the substrate binding sites facing the extracellular milieu (11, 12), and an inward facing conformation, in which the substrate binding site is shifted ~18 Å toward the cytoplasm (13). More recently, a new crystal structure has been solved, capturing a smaller inwardly oriented movement of the core domain of about 3.5 Å, which was designated as an iOFS (14). This inward movement alters the conformation of the cytoplasmic part of TM7 and generates a cavity lined by conserved residues in TM2 and TM5, leading the authors to propose that the transition to the iOFS may be linked to anion permeation (14). Because the structural data suggest that the region of TM7 where Arg-388 is located may move inwardly to open a cavity, one explanation for the effects we observe is that the negatively charged substitutions at Arg-388 stabilize the protein in an iOFS-like conformation that results in a constitutively open state. To test this hypothesis, we exploited a conformationally sensitive residue in HP1, L376C, which is accessible to NEM only under conditions favoring the inward facing conformation of the transporter (Fig. 8C) (47). The increased cytoplasmic accessibility of the L376C in the double mutant (R388D/L376C; Fig. 8D) suggests that the mutated carriers exist preferentially in conformations where TM7 is more inwardly oriented but still in an outward facing conformation. This conformation appears similar to the iOFS (14), because it retains the capacity to bind substrates and inhibitors applied extracellularly. Our observation that dl-TBOA can reduce the macroscopic anion currents in the R388D or R388E mutants to background levels further supports this idea. Although it may be counterintuitive that the anion conductance in the mutants is substrate-insensitive but sensitive to dl-TBOA, this observation actually provides support for the idea that the open channel states are closely related to the iOFS. Based on crystallographic (12) and electron paramagnetic resonance (EPR) data (55), dl-TBOA binding appears to prevent the closure of HP2, therefore restricting the carriers to an outward conformation (12, 55). We hypothesize that because dl-TBOA binding precludes subsequent conformational changes, it can effectively inhibit channel activity in the WT and the mutant transporters (R388D and R388E) by preventing them from reaching the intermediate conformational states where channel opening occurs. Recent work from Fahlke's group elegantly combined molecular dynamics simulations with electrophysiological recordings and fluorescence spectroscopy to show that the anion-conducting states can be reached through small transitions from the intermediate states of the protein (35), consistent with the close association between the iOFS and the open channel conformation that we propose here. Interestingly, their results conflict with a previously proposed hypothesis that Ser-103 (Ser-65 in GltPh) and residues nearby form part of the permeation pathway (28, 54, 56). Instead, their findings more strongly support the possibility that Ser-103, a residue in close proximity to Arg-388 (<3 Å) (11), may be involved in anion channel gating.
A number of reports have provided evidence indicating that substrate transport and substrate-gated anion conductance in EAATs are mediated by two distinct mechanisms. Point mutations at several residues have been reported to affect anion permeation without affecting substrate transport (28, 29, 56). On the other hand, sulfhydryl modifications of several introduced cysteine mutants in the transport domain inhibit substrate translocation but preserve anion permeation (34, 48, 57). Although these findings support the idea that transport and channel activities are mediated by independent mechanisms, there are also data (including the findings reported here) that support the existence of an intrinsic structural link between conformational changes associated with both functions (32, 33, 58). Here we show that the conversion of Arg-388 in EAAT1, as well as Arg-410 in EAAT4, to a negatively charged residue disrupts gating by impeding channel closure and shifting the carrier into conformations that favor anion flux rather than substrate transport. The mutant carriers predominantly exist in anion-conducting states, which resemble an iOFS, and rarely transition to inward conformations to complete the transport cycle. These findings are consistent with the idea that transport and anion-conducting states may be mediated by closely linked alternative conformations and that structural modifications that alter the equilibrium between these two states determine the time the carrier spends in conformations associated with each function. This is supported further by the observation that carrier isoforms that display much lower transport activity have comparable or even larger macroscopic anion currents when compared with those with high transport rates, despite their similar unitary current amplitudes (37). Our conclusion is also supported by work from Machtens et al. (35), which suggests that anion channel opening occurs as a branching reaction from intermediate conformations yet remains tightly coupled to transitions within the transport cycle. Thus, the work presented here fits well with the idea of a direct structural link between the conformational changes required for substrate transport and the transitions that allow channel opening and closing.
D. T. S. and S. G. A. conceived and coordinated the study and wrote the paper. D. T. S., J. J., and S. G. A. designed the experiments. D. T. S. performed the experiments. J. G. O. and J. J. designed and performed the experiments shown in Figs. 2 and and7.7. C. B. D. provided input on experimental design and assistance with the mutants. All authors reviewed the results and approved the final version of the manuscript.
We thank the National Institutes of Health Fellows Editorial Board. We thank Drs. Miguel Holmgren, Ariela Vergara-Jaque, Horacio Poblete, Abigail Nash, Spencer Watts, and Jeffrey R. Comer for thoughtful comments on the manuscript. We thank Geeta Strange and Socorro Vigil-Scott for excellent administrative support. We also thank other members of the Amara laboratory for input.
*This work was supported, in whole or in part, by National Institutes of Health Grant MH080726 (to S. G. A.) and National Institute of Mental Health intramural research program Grant MH002946. This work was also supported by American Heart Association postdoctoral fellowship 09POST2010115 (to D. T. S.). The authors declare that they have no conflicts of interest with the contents of this article.
5The abbreviations used are: